CN116324656A - Monitoring of a converter - Google Patents

Abstract

In a method for monitoring a power converter (1), a plurality of warning messages (2) are used in order to determine an error condition of the power converter (1). Wherein the current transformer has at least two types of warning messages, a first warning message type having a warning message which is dependent on the type of the current transformer, and a second warning message type having a warning message which can be defined by a user, wherein a combined evaluation of the respective first type and the respective second type of warning is used for monitoring, wherein the different types of warning messages (2) are grouped, wherein the grouping of warning messages (2) corresponds to an element of the current transformer (1).

Description

Converter Monitoring

technical field

The invention relates to the monitoring of a converter or the monitoring of a drive, wherein the drive has at least one converter, wherein the drive can also have an electric machine, such as an electric motor or a generator, in addition to the converter.

Background technique

A variable-speed operation of an electric machine (for example an electric motor) can be realized by means of a converter. When an electrical machine is used as a generator, a converter can also be used to convert the electrical current. In this way, for example, a grid feed-in can be realized. For example, for electric motor applications, grid power with constant frequency and voltage is converted to power with variable frequency and voltage.

A converter is, for example, an inverter, a rectifier or a power converter. The converter can be water-cooled and/or air-cooled. For example, converters are used in applications that place high demands on reliability and quality. Examples of converter applications include:

·Industrial pumps and fans

Oil and gas pumps and compressors, such as electric submersible pumps and high-speed compressors

· Boiler fans for power generation (induced and forced draft)

· Cleaning of water pumps and waste water pumps

multi-motor applications and synchronous transmissions (e.g. pipelines in the oil and gas industry),

·etc.

These application examples generally involve the use of converters, where especially high power is required. This is especially for powers in the single-digit, double-digit or triple-digit megawatt range. For this purpose, a medium-voltage converter is preferably used. These can be described as medium voltage converters. Voltages greater than or equal to 1000V can be regarded as medium voltage. A voltage of 4000V or 6000V may also be referred to as medium voltage. In addition to power semiconductors, converters also have regulators or controllers, such as transformers and/or choke coils. The converter is, for example, a variable frequency drive (VFD).

If there is a problem with the converter, it can be identified and recorded or saved in a log file. A log file represents a log in which identified and logged issues are log entries. These log entries can be alerting users to potentially critical events. These log entries can also be errors, ie error messages, in order to notify the user of the error or to log the error. For example, key events are also errors. Therefore error or warning messages are generated for the event. For example, an error message can also be a disturbance report. So in the presence of errors, especially in the presence of interference. Errors can cause or have caused the converter and/or drive to malfunction. For example, if there is a problem with the drive, the identified problem is logged in a log entry. These log entries can be warnings (to alert the user of potentially critical events) or errors (logging an error that caused the drive to fail).

Warnings indicating a sub-optimal state of the drive are currently sent from the drive or converter to a UI (User Interface), eg accessible via the drive's or converter's HMI and/or via a cloud solution. One or more warnings are then analyzed by the corresponding drive specialist and measures are taken and implemented by the specialist.

Contents of the invention

An object of the invention is to improve the monitoring of a converter or a drive.

A solution to this object is given by a method according to claim 1 or by an analysis system according to claim 11 . For example, some designs are given according to claims 2 to 10.

In the method for converter monitoring, a number of warning messages are used in order to derive error conditions of the converter. Here and below, in addition to the converter, the method also involves a drive, which has a converter. In addition to the converter, the drive can also have a transmission and/or an electric machine. Inference of error cases also involves prediction of error cases. Therefore, prediction of error cases involves situations where the error has not yet occurred but will occur with some probability after prediction. In this case, errors also refer in particular to disturbances in the converter or drive.

This method for monitoring of converters uses eg log data from the converters.

In a method for monitoring a converter, a number of warning messages are used to deduce an error situation of the converter, wherein the converter has at least two types of warning messages, a first warning message type which has a function depending on warning messages of the type of converter, and a second warning message type with warning messages definable by the user, wherein for the monitoring of the converter a combination of a corresponding first type and a corresponding second type of warning is used The evaluation of , where different types of warning messages are grouped, where the grouping of warning messages corresponds to the elements of the converter. Therefore, the converter can have at least two types of warnings, a first warning type, which depends on the type of converter, ie can be determined in particular by this type, and a second warning type, wherein the second warning type has User-definable, ie user-determinable, warnings, wherein for event monitoring it is also possible to use a combined evaluation of errors or warnings of the respective first type and of the respective second type. The converter can thus be designed, for example, by the user in such a way that the user defines individual messages for the converter. Examples of reports are errors or warnings, that is, error messages or warning messages. With personalized messages, it is possible to create personalized event monitoring that relies on user personally created messages. This improves event monitoring and enables it to be more accurate. Messages defined by the user are for example stored in SOPs (System Operating Procedures) or defined there. Such user-defined messages can be flagged when the report is displayed. For example, user-defined messages are based on signals from I/O interfaces present in the installation integrating the converter. For example, the signal on which the user-defined report is based can be a digital signal or an analog signal. For example, user-defined messages can be generated based on a single signal and/or based on a combination of signals. Such signals can relate, for example, to an emergency stop, opening of a door, blown fuse, undervoltage, overvoltage, fault current, overcurrent, fan failure, failure of a power module of the converter, bypassing of a power module of the converter, Insulation errors, insulation warnings, communication errors, especially of the power modules of the converter, errors or warnings of converter cooling, converter precharging, etc. It can be seen that the messages created individually for the converters involved in individual use (industrial installations) are important. These messages individually created by the user can advantageously be used for event monitoring of the converter in its individual environment, ie an industrial facility. This improves the quality of monitoring. In this case, the user of the converter is, for example, a person who operates the converter. This operation can be performed, for example, by an operator of the converter, a commissioning person, or the like.

In one refinement of the method, the warning message, ie the warning, can relate, for example, to the following subjects: cooling body temperature too low, cooling body temperature too high, current asymmetry, temperature too high in the converter, line reactor temperature near shutdown limit, cabinet heating failure, motor heating failure, motor winding overtemperature, insulation failure, cabinet fan failure, low speed, motor lock and/or control voltage, etc. Such warning messages can be individually created for the converter specific to the installation. Depending on the installation, different requirements can be placed on the converter. The cooling of the converters can also be facility specific, so separate warnings are created for the converters in a specific facility environment, and these warnings are then combined into groups. In such a group there can also be at least one warning message, which depends on the type of converter. Therefore, a separate prediction of feasible error is feasible, which error is facility-specific.

In one refinement of the method, the warning message of the second type is dependent on the installation in which the converter is integrated. Such dependencies can for example relate to at least one of the following: external cooling of the converter, external temperature sensors, external humidity sensors, external safety sensors (e.g. door switches), external voltage monitoring of the power supply of the controller, sensors, etc.

In one refinement of the method, the warning message of the second type is dependent on a combination of signals. This combination can be selected by the user. The combination specifically involves Boolean connections such as AND, OR, NOT. Examples of signals relate in particular to the following reports: voltage supply of power electronics, voltage supply of control and/or regulation electronics, door opening, door closing, etc. Thus, the user is able to design individual warnings.

In one refinement of the method, faults of the converter can be predicted by algorithmic evaluation of warnings (warning messages). Warnings contain not only quantitatively measurable information (e.g. via threshold warnings), but also qualitative information about the state of the converter or drive. For this, algorithms can be used.

In one refinement of the method, the warning messages are grouped, wherein the groups of warning messages correspond to the components of the converter. Thus, for example, the following warning groups are possible: transformer, cooling, bypass, power semiconductors, battery, electric machine, transmission and/or bearing. These groups can be linked to each other in the analysis of warning messages. Warning messages from different groups can depend on each other.

In one refinement of the method, warning messages can be retrieved, for example, from log data. When considering warnings, in one design, for example, a certain time period and/or a time period after a converter restart can be considered. In this way, for example, it is possible to use or take into account only relevant warning messages which are relevant in terms of time.

In one refinement of the method, one or more warning messages are identified, ie marked with, for example, a time-sequence-related identifier, ie marked. For example, time series are generated by timestamps, which are linked with warning messages (and vice versa). However, the identification, ie the flag, of the warning message can also be, for example, the degree of urgency of the warning message, the source of the report and/or the name (description) of the warning message, etc.

In one configuration of the method, erroneous inferences involve predictions of erroneous situations. An error inference means that an error condition is expected based on one or more warning messages (in combination or individually) if no countermeasures are taken. In an advantageous refinement, the prediction of error cases is also linked to the time specification. A time specification provides, for example, an answer as to when the occurrence of an error can be counted with a certain probability from a specific point in time (eg withdrawal of a warning message or invocation of a warning message). Thus, it can be concluded, for example, that the probability of error occurrence increases by a certain value as the waiting time increases.

In one refinement of the method, at least one rule is used for monitoring, the rule relating to the occurrence of one or more warnings from one or more groups. For example, the occurrence of one or more specific errors in one or more groups can be correlated and analyzed.

In one refinement of the method, the evaluation of the warning message generates a message for an impending error situation, ie in particular for an impending disturbance. Therefore, the user of the converter or drive can take countermeasures. In one refinement of the method, countermeasures of this type are derived analytically and are initiated, in particular automatically.

In one configuration of the method, the severity of the impending error is derived. Then, it is better to take appropriate action according to the severity. This is, for example, planning a quick repair or ordering spare parts, which may be time-consuming.

In one refinement of the method, a large number of warning messages can be converted into messages for impending error situations. In addition to timely anticipation of impending error situations, this has advantages, for example, in being able to reduce the number of user-relevant messages. This improves in particular the overview and easy handling of the converter or drive. Messages of impending error situations that can preferably be avoided can be abbreviated eg Precog. For example, this concept can be understood as "pre-identification".

In one design of the method, artificial intelligence is used. According to this method, artificial intelligence can also be trained. Therefore, the knowledge of experts can be replaced by artificial intelligence.

In one refinement of the method, a message is generated regarding the temporal occurrence of the error situation. For example, it is also possible to analyze what further consequences the occurrence of an error has. Thus, for example, it can be analyzed whether it is advantageous to react quickly or slowly to warning messages or messages about impending errors (Precog). This can depend, for example, on expected downtimes.

It is advantageous to react quickly or slowly. 8. A method according to any one of claims 1 to 7, wherein a message is generated to prevent an error situation.

In one refinement of the method, the warning message is received via the Internet. Thus, for example, the analysis device can receive reports from the converter or drive via the Internet. In particular, after the analysis, the analysis means can send data related to the analysis result to a user interface (UI: User Interface). For example, such data can be provided to the user interface via an Internet connection. For example, the user interface is located where the converter or drive is located, in the console, etc.

In addition to the method, the invention also relates to an analysis device or an analysis system having an analysis device. The analytical device or analytical system is especially designed for carrying out the method. For example, an analytical system can also be called a pre-error system because it is able to predict potential errors before they occur.

The evaluation system of the converter or the drive especially has artificial intelligence, wherein the evaluation system is in particular spatially separated from the converter.

In one configuration of the method or in one configuration of the analysis system, there is a digital platform for optimizing drive systems, motors, converters, etc., wherein reports or log files are collected and/or transmitted. Thus, data can be transmitted from the converter to the analysis system via the Internet, ie via the cloud. Such data, such as logs (log data), are obtained from digital platforms and contain, for example, at least one of the following information:

- the timestamp of the log event,

- the severity of the log event: info, error or warning,

- the text of the log event,

- Actors of logging events (producers/sources of logging events).

For example, the occurrence of one or more errors after a certain error-free time can indicate a drive failure, which results in an unplanned shutdown of the drive. Errors are especially alerts with a severity of "Error". For example, a drive failure occurs when the drive triggers a shutdown due to one or more active errors. Some errors, and consequently some failures, can be predicted by specific warnings that occur before they occur.

The analysis system is specifically designed as a pre-error system and analyzes the generated warnings using a developed algorithm. Thus, the system identifies potential errors before they actually occur. In particular, if a potential failure of a drive can occur, the analysis system takes into account the severity of the alarm to predict the occurrence of the error.

In one design of the system or method, warnings are classified into coherent groups and/or specific decision rules based on expert knowledge are applied to these groups. For example, it is possible to create four pre-error groups for warnings:

- input/transformer,

- battery, control, output,

-cool down

-bypass

These groups are mentioned here as examples. The group specifically covers the critical components of the drive (in particular, all critical components) and is able to predict the failure of these critical components for a period of at least two more hours. When there is a specific warning for a specific group with specific rules, the system outputs a positive prediction of potential drive failure from the identified group related to potentially critical components of the drive.

In one refinement of the system, the groups are based on a physical relationship of warnings, wherein these warnings can be of different types, and in particular also on a historical or statistical evaluation of the operating history of the respective converter type. Therefore, grouped warnings are associated with one error type. For example, a physical relationship occurs in the cooling topic group because related alarms (warnings) such as cooling tank condition (cooling tank level), cooling liquid flow and/or cooling liquid temperature are aggregated in this group to the topic cooling.

In one configuration of the system, the system has special rules as a pre-error. These rules depend on whether a report (Precog) is generated for (potentially) impending error conditions. To this end, for example reports are removed from the history of logs and/or extended by further discoveries.

In analyzing a design of the system, use at least one of the following rules:

-Rule 1:

Each group has at least 1 alarm: Group 1, Group 2, Group 3

-Rule 2:

Each group has at least 1 alarm: Group 1, Group 2

- Rule 3:

Each group has at least 1 alarm: Group 2, Group 3

-Rule 4:

At least 3 alarms in at least one group: Group 1, Group 2, Group 3

- Rule 5;

At least 1 alarm in group 4.

In one configuration of the analyzing system, which is designed to report impending error situations, wherein the analyzing system thus represents a pre-error system, wherein the analyzing system determines messages for a specific point in time or a specific period of time, wherein , in particular only warnings for a certain period of time are considered in the calculation (e.g. warnings for the last 14 days, i.e. warning messages (example for the maximum value)). Such time parameters are able to match conditions. This time period is in particular a characteristic time, which is defined during the development of the converter or drive. This is especially useful to ensure that only relevant warnings are considered by the pre-error system. The time stamp of the acknowledged warning (acknowledgment time stamp) must not exist in the past in one configuration. Acknowledges the end of the activation phase of the time stamped warning. In one configuration of the analysis system, the confirmed time stamps are taken into account during the analysis. In the analysis, it is advantageous to consider whether the warning message was acknowledged without carrying out a measure (eg repair, replacement, etc.), or whether the warning message was acknowledged after carrying out an action. To this end, a first example can be given as follows: In the past 24 hours, two important warnings occurred one after the other. The first warning is part of group 1 and the second warning is part of group 2. When the second important warning occurs, the pre-error system is activated because Rule 2 (see above) can apply here. A second example might produce the following: Two critical alerts in a row within the past 24 hours. Both warnings are part of group 1. When the second important message comes, the pre-error system will not be activated, because here no rules can be applied Then, if another message from group 1 arrives, a message about the impending error will be generated, because the rules can be applied 4 (see above).

The analysis system makes it possible to analyze all warnings in real time during production, ie in particular during operation of the drive, without extensive technical knowledge. Manual analysis is often not performed in an adequate manner at production facilities since the ongoing manual analysis is time-consuming. The method provides automatic prediction of potentially critical conditions before they occur to prevent failures. Thus, potential downtimes of drives can be reduced by scalability with respect to all drives connected to the analysis device, for example in a production facility. As a result, at least one of the following functions can be implemented, for example:

- Ability to notify about possible drive failures in the near future,

- be able to notify why a drive failure can occur in the near future,

- Ability to create recommendations and explanations to prevent drive failures.

In one configuration of the analysis system, the analysis device can be used to implement the cloud analysis method, since the log file data from the drive is stored in the cloud. Therefore, a pre-error system is provided for scalable cloud instances in a suitable Python environment. These technical features enable process automation. The method is realized by an algorithm based on expert knowledge.

In one design of the method, the drive's logs are analyzed by an algorithm based on expert knowledge (with appropriate labels), which classifies the log warnings into different relevant groups and applies specialized decision rules to this classification, which also considers occurrence and identification/acknowledgment of these warnings.

Description of drawings

The features of the individual objects claimed or described can easily be combined with one another. In the following text, the invention is described and explained in more detail by way of example with reference to the drawings. Those skilled in the art can combine the features shown in the figures to form new embodiments without departing from the invention. Similar elements include the same reference numerals. shown in the figure

Figure 1 shows an overview of the monitoring method,

Figure 2 shows the time window of the report,

Figure 3 shows the truth matrix,

Figure 4 shows the reaction time analysis,

Figure 5 shows the statistics,

Figure 6 shows the flowchart,

Figure 7 shows the first regular graph,

Figure 8 shows the second regular graph, and

Fig. 9 shows a third regular graph.

Detailed ways

An overview of the monitoring method is shown according to FIG. 1 . Converter 1 is monitored. The converter 1 is part of an installation 6 . The installation 6 also has a communication device 7 (DSC: Drive System Connection), which enables communication of the converter 1 . Thus, messages such as a warning message (Warning) 2 can be sent to a log file (German: Logdateien). Such reports concern bugs or general events. Such reports are sent via the Internet 4 . For this purpose, for example, a workstation (DSEW: Drive System Expert Workstation) can be provided. Information about the state of health (health level) 8 can thus be sent to the interface 5 for the analysis diagram 9 of the drive. The log data (asset log) 10 is also sent to the analysis system 6 , wherein the analysis system 6 has the analysis device 3 . The analysis device 3 is provided, for example, with artificial intelligence. The evaluation device 3 determines the state of health of the installation 6 or of one of the components of the installation (for example the drive 1 ) and transmits this state of health 11 to the workstation 4 in the Internet. The model is located in Analysis System 6. The model can be improved, ie updated, by means of a data analyst (data scientist) 12 . The data analyst 12 can provide the new error 13 to the expert 13 . The expert 13 can return evaluated errors 14 (marked as faults). For example, the expert 13 receives an email 15 when a future error (pre-failure) 15 has been drawn. The expert 13 is also able to transmit information about potential errors 17 to the user 16 via a user interface (UI).

The diagram according to FIG. 2 shows a time window 18 in which a state of health (degree of health) 20 is plotted over time 19 . There is a healthy state 21 , a warning state 22 , an error about to be recognized (pre-failure or predictive) state 23 and an error state 24 . Health also decreases in that order. A time window 18 can be guided over these states, wherein a time line 25 for the warning is given depending on these states.

The illustration according to FIG. 3 shows a truth matrix (confusion matrix) 26 . It shows how predictions including real events can be flagged, ie identified.

The illustration according to FIG. 4 shows the result of the reaction time 27 until an error 24 occurs.

The diagram according to FIG. 5 shows the statistical prediction accuracy of the model for predictable errors. Here, the distributions predicted by the air-cooled, and water-cooled models are shown.

The illustration according to FIG. 6 shows a flow diagram similar to that of FIG. 1 . The data (asset log) of the converter 1 is directed to an analysis system 6 via a fleet management server 4 representing the Internet. The analysis system 6 carries out the derivation of the early detection of errors and can therefore also be referred to as a pre-error system. The analysis system 6 outputs a corresponding pre-error report (pre-failure) 30 .

The illustrations according to FIGS. 7 , 8 and 9 show different rule graphs. Fig. 7 shows the first rule. Fig. 8 shows the second rule. Fig. 9 shows the third rule. Areas 31, 32 and 33 are formed for different events, respectively. Area 31 relates to groups 34 , 35 , 36 and 37 in which warnings are grouped. In the first group 34 there are grouped warnings related to the input and the transformer. In a second group 35 there are grouped messages, regulation and outputs related to the batteries of the converter. In a third group 36 there are messages related to cooling grouped. In a fourth group 37 there are grouped messages of warnings related to bypassing of power semiconductors in the converter. According to a first rule, there must be at least one warning in groups 35 , 35 and 36 to be active, ie to trigger a pre-error warning ( Precog1 ) 38 . In this way, error 41 can be prevented. As shown in the illustrations in Figures 7, 8 and 9, there are three different messages 38, 39 and 40 in area 32 for warning of impending errors (Precog 1, 2 and 3). In the area 33 for errors, there are three different errors 41 , 42 and 43 to be predicted or prevented. The illustration according to FIG. 8 shows a second rule according to which at least three warnings must be present in the third group 36 in order to activate pre-error reporting (Precog 2) 39 in order to be able to predict or prevent errors 42 . The diagram according to FIG. 9 shows a third rule according to which at least one warning must be present in the fourth group 37 in order to activate pre-error reporting (Precog 3 ) 40 in order to be able to predict or prevent errors 43 .

Claims (11)

1. A method for monitoring a power converter (1), wherein a plurality of warning messages (2) are used in order to deduce an error situation of the power converter (1), wherein the power converter has At least two types of warning messages, a first warning message type with warning messages depending on the type of converter and a second warning message type with warning messages which can be defined by the user, wherein the corresponding first warning message type is used for monitoring Evaluation of a combination of warnings of a type and a corresponding second type, wherein warning messages (2) of different types are grouped, wherein the grouping of warning messages (2) corresponds to the components of the converter (1). 2. A method according to claim 1, wherein said second type of warning message depends on the installation in which said converter (1) is integrated. 3. The method according to claim 1 or 2, wherein warning messages are marked, wherein said marking relates to a time series. 4. A method according to any one of claims 1 to 3, wherein an erroneous inference involves a prediction of the erroneous situation. 5. A method according to any one of claims 1 to 4, wherein for monitoring purposes rules are used, wherein the rules relate to the occurrence of one or more warnings in one or more groups. 6. A method according to any one of claims 1 to 5, wherein artificial intelligence is used and/or trained. 7. The method as claimed in any one of claims 1 to 6, wherein a message is generated regarding the temporal occurrence of the error situation. 8. A method according to any one of claims 1 to 7, wherein a message is generated to prevent the error condition. 9. The method according to any one of claims 1 to 8, wherein a warning message is received by the analysis device (3) via the Internet (4), wherein the analysis result is sent to the user interface (5). 10. The method according to any one of claims 1 to 9, wherein an analysis system (6) according to claim 11 is used. 11. An evaluation system for a converter (1), in particular with artificial intelligence, wherein the evaluation system (6) is spatially separated from the converter (1).

Images